The big model industry is in the ascendant, and A-share companies are scrambling to get involved.
Our reporter Kang Xi
The rapid development of artificial intelligence technology with large model as the core has become a new engine to promote economic development. Recently, Tianrongxin’s self-developed Tianwen model has successfully passed the "Generative Artificial Intelligence Service Filing". According to the Ministry of Industry and Information Technology, there are nearly 200 large models of generative artificial intelligence services that have been filed and launched to provide services to the public in China.
According to industry insiders, in the future, with the accelerated development of large-scale model technology, the accelerated application of technology enterprises and the enhanced government support, the development of China’s large-scale model industry will enter the fast lane. From the technical point of view, AI (artificial intelligence) big model is developing in the direction of multi-modal, specialization, generalization and cooperation of large and small models.
Broad market prospects
In 2024, the government work report put forward the action of "Artificial Intelligence+"for the first time, which gave guidance for the large model to be deeply embedded in thousands of industries. Large-scale model is an important technical achievement in the field of artificial intelligence, and it is reshaping the development model of traditional industries.
In the past year, China’s large-scale models have made great progress, and a large number of high-quality large-scale models have blossomed. According to the Ministry of Industry and Information Technology, there are more than 4,500 artificial intelligence enterprises in China, and nearly 200 large models of generative artificial intelligence services have been completed and put on line to provide services to the public, with more than 600 million registered users.
In order to promote the healthy development of AI big model industry, China has issued a series of supporting policies and plans, which provide a strong guarantee for industrial development. In January this year, the "Implementation Opinions on Promoting the Future Industrial Innovation and Development" jointly issued by seven departments including the Ministry of Industry and Information Technology proposed to accelerate the breakthrough of computing technology innovation such as quantum and photon, accelerate the deep empowerment of brain-like brains, swarm brains and large models, and accelerate the cultivation of intelligent industries; In May, the Central Network Information Office and other departments jointly issued the "Information Standards Construction Action Plan (2024-2027)", which clearly stated that it is necessary to accelerate the development of large-scale models and generative artificial intelligence standards.
According to the Research Report on China AI Big Model Industry Market Prospect Forecast in 2024 issued by China Commercial Industry Research Institute, the market size of China Big Model Industry has increased from 1.5 billion yuan in 2020 to 7 billion yuan in 2022, with an average compound annual growth rate of 116.02%. Analysts of China Commercial Industry Research Institute predict that the scale of large model industry in China will reach 21.6 billion yuan in 2024.
Xiong Wei, vice president of Huaxin Research Institute of China Gongxin Publishing Media Group, said that the global AI big model market is increasingly competitive. With its deep accumulation and strong technical strength in the field of artificial intelligence, the technology giants occupy a dominant position in the market. At the same time, innovative SMEs are also actively investing in the research and development and application of AI big models. These enterprises usually have unique technical routes and application scenarios, and can achieve breakthroughs and innovations in specific fields. With the continuous progress of technology and the expansion of application scenarios, these small and medium-sized enterprises are expected to occupy a larger share in the market.
From the technical path, Xiong Wei believes that the artificial intelligence big model is developing in the direction of multi-modal, specialization, generalization and cooperation of large and small models. Multi-modal large model can handle multi-modal data and realize more complex intelligent tasks; The specialization model is customized and optimized for the needs of specific industries or fields; The generalized grand model aims to realize cross-domain intelligent application; The cooperation of large and small models is one of the important directions for the development of large AI models in the future.
All localities are scrambling to increase investment.
The rapid development of artificial intelligence big model industry has become the new engine of digital economy growth, and all localities are seizing strategic opportunities, embracing big models in an all-round way and competing for industrial layout.
At present, China has formed three industrial clusters and development zones of artificial intelligence in Beijing-Tianjin-Hebei, Yangtze River Delta and Pearl River Delta, and a number of industry leaders and unicorn enterprises have emerged to promote the rapid development of China’s artificial intelligence industrial clusters. Among them, Beijing has outstanding advantages in the country.
In July this year, the Beijing Municipal Development and Reform Commission and others jointly issued the Action Plan for Promoting "Artificial Intelligence+"(2024-2025), which proposed that by the end of 2025, efforts should be made to form 3-5 advanced, available and independently controllable basic large-scale model products, 100 excellent large-scale model products and 1,000 successful cases in the industry.
Chengdu is accelerating the construction of a new generation of national artificial intelligence innovation and development pilot zone and a national pilot zone for artificial intelligence innovation and application, promoting large models to empower thousands of industries and creating fertile land for the development of artificial intelligence industry.
On October 19th, at the kick-off meeting of the First National "Artificial Intelligence+"Industry Application Innovation Competition held in Chengdu, the data disclosed by Chengdu Economic and Information Bureau showed that in the first three quarters of 2024, the scale of artificial intelligence industry in Chengdu exceeded 80 billion yuan, with a growth rate of over 30%.
Chengdu strengthens algorithm iteration with applied thinking, promotes large-scale models to empower thousands of industries, and focuses on building influential large-scale models of explosions in China. More than 30 large-scale model development and application enterprises have been cultivated, and more than 40 large-scale models with industry influence have been developed, covering many industries such as industrial monitoring, digital cultural creation, medical and health care, and 7 of them have been filed or registered through national generative artificial intelligence services.
The "Several Measures to Promote the Innovation and Development of Artificial Intelligence Large Models in Shanghai (2023-2025)" clearly States that it will focus on supporting the construction of demonstration application scenarios in the fields of intelligent manufacturing, biomedicine, integrated circuits, intelligent education and teaching, technology and finance, design creativity, autonomous driving, robots, digital government, etc., and creating benchmark large model products and services. Promote the development of large models and related fields in all directions, further strengthen the artificial intelligence industry fund, better guide the participation of social capital, and increase the input of resources to the artificial intelligence industry in Shanghai.
Positive layout of listed companies
The big model industry has shown a vigorous development trend. Facing this blue ocean market, A-share companies have increased their investment and rushed into the big model industry.
The new generation cognitive intelligence model "iFLYTEK Spark" launched by Iflytek is continuing to empower thousands of industries, accelerate the application, and join hands with leading enterprises in many fields to build a benchmark for the application of large models. Iflytek’s third quarterly report shows that, according to relevant statistics, from January to September, the number and amount of winning bids for large-scale model-related projects in Iflytek are in the leading position in the industry.
Mango Super Media’s self-developed "Mango" model has been applied in more than 60 business scenarios such as script production, high-definition production and broadcasting, and post-production. The "Mango" big model will also make a breakthrough in the direction of video generation, and will be applied to the production of micro-short plays and interactive products with C-end content as soon as possible.
Wanxing Technology attaches great importance to the localization content generation ability of AI technologies such as large models, and its Wanxing "Skyscreen" large model has been approved by the Interim Measures for the Management of Generative Artificial Intelligence Services issued by the National Network Information Office and the National Development and Reform Commission, and has been commercialized on a large scale.
Shanghai Ganglian has developed a vertical large language model "Ganglian Master" in the commodity industry, and developed the application of "Xiaogang" digital intelligent assistant with the help of the ability of "Ganglian Master".
A wholly-owned subsidiary of Pingzhi Information signed a strategic cooperation agreement with Beijing Zhipu Huazhang Technology Co., Ltd., which stipulated that the two sides should jointly study the application of multi-modal large-scale models, explore the integration of deep learning, robots and visual languages, empower large-scale model developers with large models, and jointly explore the MaaS market.
According to industry insiders, grasping the application demand is one of the breakthrough directions of big model related enterprises. The general intelligence of the big model must be refined into various industries, and it must be fed with professional data in the industry. Only by training can it form a scene-based, customized and personalized proprietary model, can it bring artificial intelligence revolution to various vertical fields.
Reporting/feedback